© 1999 Elsevier Science Ltd. All rights reserved. For article fee, see p. IV.TCB, Vol. 9, N¼ 12 (0962-8924) TIBS, Vol. 24, N¼ 12 (0968-0004) TIG, Vol. 15, N¼ 12 (0168-9525)PII: S0962-8924(99)01667-0 PII: S0968-0004(99)01479-6 PII: S0168-9525(99)01891-0Even a cursory consideration of metazoan anatomy anddevelopment forces the realization that the associations ofcells in epithelia, their attachment to basement membranesand the migrations of cells and projections of neurons all requireselective adhesion of cells to one another and to extracellularmatrices (ECMs). Recognition of this requirement led to a spiriteddebate between proponents of a large number of highly selectiveadhesion receptors, and advocates of models in which quantita-tive differences in adhesive strength, without necessarily a largespectrum of individual specificities, were invoked to explain dif-ferential cell adhesion. Similarly, the phenomenon of induction,in which one tissue influences the developmental fate of adjacenttissues, clearly relies on cell–cell interactions, and experimentalembryologists attempted to define whether induction relies ondiffusible signals or on cell–cell or cell–matrix contacts. Neitherthe issue of specificity of cell adhesion nor the question of themechanistic bases of induction could be resolved without mol-ecular biology. Now, with the benefit of a couple of decades ofmolecular analysis, we can see that there is some truth to all of theearlier models. The specificity of cell adhesion comes from com-binatorial expression and interactions among a large, but notunlimited, number of adhesion receptors, and induction relies ondiffusible ligands binding to receptors, on cell–cell contacts andon cell–matrix adhesion. The distinctions among these threemechanisms are not actually that great – adhesion receptors signal much like receptors for growth factors and should be considered in parallel with them.Before considering the biological functions of cell adhesion, weneed to define the players. Figure I in Box 1 diagrams the struc-tures of representative cell–cell adhesion receptors. Fortunately,many adhesion receptors fall into a relatively small number offamilies, the major ones being shown in Fig. I. Other families ofadhesion receptors, such as syndecans and other membrane-bound proteoglycans, the disintegrin family and others are lesswell understood at this time. In addition to their roles in bindingcells to their neighbours (Fig. I) or to ECM (Fig. 1), engagementof cell-adhesion receptors has major effects on many aspects ofcell behaviour – cell shape and polarization, cytoskeletal organiz-ation, cell motility, proliferation, survival and differentiation.How do they accomplish all these functions?Cytoskeletal connectionsCrucial to the effects of adhesion receptors on intracellularorganization and cell motility is the fact that their cytoplasmicdomains connect to the cytoskeleton. Figure 1a shows how integrins bind to linker proteins, which in turn make direct and indirect connections to F-actin filaments, thus establishing a mechanical link between the fibrils of the ECM and the filaments of the cytoskeleton9,10. The connection of classic cadherins to the actin cytoskeleton that occurs at cell–cell junc-tions is analogous, although the molecules involved are different(Fig. 2a)1,11,12. Although integrins appear to be the major recep-tors for ECM, they are not the only ones. One well-studiedexample, of considerable interest because of its involvement inMillennium issueRichard O. [email protected] HughesMedical Instituteand Center forCancer Research,Dept of Biology,MassachusettsInstitute ofTechnology,Cambridge, MA 02139, USA.Cell adhesion: old and new questionsRichard O. HynesMetazoans clearly need cell adhesion to hold themselves together, but adhesion does much more than that.Adhesion receptors make transmembrane connections, linking extracellular matrix and adjacent cells to theintracellular cytoskeleton, and they also serve as signal transducers. In this article, I briefly summarize ourpresent understanding of the molecular basis and biological consequences of cell adhesion and discuss how ourcurrent knowledge sheds light on questions of specificity of cell adhesion. I offer some thoughts and speculationsabout the evolution of cell-adhesion molecules and processes, consider their inter-relationships with other forms ofcellÑcell communication and discuss unresolved questions ripe for investigation as we enter the postgenomic era.Regis B. Kelly ¥ Membrane traffickingM33U. S. A. 95, 11199Ñ1120424 Takei, K. et al. (1998) Generation of coatedintermediates of clathrin-mediatedendocytosis on protein-free liposomes. Cell94, 131Ñ14125 Matsuoka, K. et al. (1998) Coat assemblydirects v-SNARE concentration into syntheticCOPII vesicles. Mol. Cell 2, 703Ñ70826 Faundez, F. et al. (1998) A function for theAP3 coat complex in synaptic vesicleformation from endosomes. Cell 93, 423Ñ43227 Owen, D.J. and Evans, P.R. (1998) Astructural explanation for the recognition oftyrosine-based endocytotic signals. Science282, 1327Ñ133228 Blagoveschenskaya, A.D. et al. (1999) A complex web of signal-dependenttrafficking underlies the triorganellardistribution of P-selectin in neuroendocrinePC12 cells. J. Cell Biol. 145, 1419Ñ143329 Chen, C.S. et al. (1997) Changes in the spectralproperties of a plasma membrane lipid analogduring the first seconds of endocytosis inliving cells. Biophys. J. 72, 37Ñ5030 Mukherjee, S. et al. (1999) Endocytic sortingof lipid analogues differing solely in thechemistry of their hydrophobic tails. J. CellBiol. 144, 1271Ñ128431 Simons, K. and Ikonen, E. (1997) Functionalrafts in cell membranes. Nature 387, 569Ñ57232 Kuehn, M.J. et al. (1998) COPIIÐcargointeractions direct protein sorting into ER-derived transport vesicles. Nature 391,187Ñ19033 Grote, E. et al. (1995) A targeting signal inVAMP regulating transport to synapticvesicles. Cell 81, 581Ñ58934 Salem, N. et al. (1998) A v-SNARE participatesin synaptic vesicle formation mediated by theAP3 adaptor complex. Nat. Neurosci. 1, 551Ñ55635 Springer, S. and Schekman, R. (1998)Nucleation of COPII vesicular coat complex byendoplasmic reticulum to Golgi vesicleSNAREs. Science 281, 698Ñ70036 Springer, S. et al. (1999) A primer on vesiclebudding. Cell 97, 145Ñ14837 Collins, J. and Mayblin, B. (1997) IntroducingDerrida, Totem Booksmuscular dystrophies, is the dystroglycan complex, which con-nects